library(tidyverse)
library(reshape2)
library(rmarkdown)

Select predictors for our dataset and have one dataset that is groupped by states.

df_com is by communities df_state is by states

df <- read_csv("crime.csv")
df_com <- select(df, communityName, state, ViolentCrimesPerPop, population, racepctblack, racePctWhite, racePctAsian, racePctHisp, agePct12t21, agePct12t29, agePct16t24, agePct65up, medIncome, PctPopUnderPov, PctNotHSGrad, PctUnemployed, TotalPctDiv, PctKidsBornNeverMar, RentMedian, PctForeignBorn )

df_state_dec <- df_com |> group_by(state) |> summarise(ViolentCrimesPerPop = sum(ViolentCrimesPerPop), population = sum(population), racepctblack = mean(racepctblack), racePctWhite = mean(racePctWhite), racePctAsian = mean(racePctAsian), racePctHisp = mean(racePctHisp), agePct12t21 = mean(agePct12t21), agePct12t29 = mean(agePct12t29), agePct16t24 = mean(agePct16t24), agePct65up = mean(agePct65up), medIncome = mean(medIncome), PctPopUnderPov = mean(PctPopUnderPov), PctNotHSGrad = mean(PctNotHSGrad), PctUnemployed = mean(PctUnemployed), TotalPctDiv = mean(TotalPctDiv), PctKidsBornNeverMar = mean(PctKidsBornNeverMar), RentMedian = mean(RentMedian), PctForeignBorn = mean(PctForeignBorn))

df_state <- df_state_dec %>%  mutate_if(is.numeric, round, digits=2)

paged_table(df_com)

Residuals vs Fitted Values of all predictors

predictors <- df_com[,4:20]

# Title {.tabset .tabset-fade}
scatter <- function(input) {
  x <- unlist(input)
  y <- df_com$ViolentCrimesPerPop
  fit <- lm(y ~ x, data = df_com)
  plot_title <- paste("Scatter Plot of", names(input), "vs. ViolentCrimesPerPop")
  assign(paste0("plot_var", names(input)), plot(fit, 1, main = plot_title))
}

# Check for linearity with residuals vs fitted values
for (i in 1:length(predictors)) {
  scatter(predictors[, i])
}

Boxplots

df_box <- df_com[, c("racepctblack", "racePctWhite", "racePctAsian", "racePctHisp", "ViolentCrimesPerPop")]
df_box_melt <- melt(df_box, id.vars = "ViolentCrimesPerPop", variable.name = "Race")

ggplot(df_box_melt, aes(y = value, fill = Race)) +
  geom_boxplot() + ylab("Percentage of people of each race")

df_box_state <- df_state[, c("racepctblack", "racePctWhite", "racePctAsian", "racePctHisp", "ViolentCrimesPerPop")]
df_box_melt_state <- melt(df_box_state, id.vars = "ViolentCrimesPerPop", variable.name = "Race")

ggplot(df_box_melt_state, aes(y = value, fill = Race)) +
  geom_boxplot() + ylab("Percentage of people of each race")

ggplot(df_com, aes(x =  ))

ggplot(df_state,aes(x = medIncome , y= ViolentCrimesPerPop, label = state)) + geom_point(aes(color = state)) + geom_text() + theme(legend.position = "none")

Density graph of ViolentCrimesPerPop in percentages

ggplot(data = df_com, aes(x = ViolentCrimesPerPop)) + geom_density() + scale_y_continuous(labels = scales::percent_format()) + labs(title = "Density graph of ViolentCrimesPerPop in percentages")